The Conundrum of Identifying the Poor

I used to think that giving aid was easy. You just find those in need and give them money, incentives or beneficial programs, right? It turns out that even the seemingly simple initial process of identifying the poor is not as easy as it sounds. Not only has a truly efficient method of identifying those living in poverty yet to be established, but there are discrepancies between community satisfaction with known methods and the method’s official success rate.

The main challenge that is faced by researchers and potential benefactors in developing countries when identifying poor people is a lack of reliable income data. Many of the poorest people work informally and/or inconsistently, with few or no verifiable income records. Considerable and creative efforts therefore need to be made to identify intended beneficiaries if aid money is not to be misdirected toward wealthier households.

To date, Proxy Means Tests (PMTs) have proved the most accurate method of identifying families living on USD$2 or less a day. This method uses a wide range of household asset indicators to determine a household’s quality of living (in a material sense). For example, does the household have a television, cooker, beds etc? Depending on the implementers, PMTs may also include data concerning education, composition and occupation.

In terms of popularity and accuracy PMTs are closely followed by the Community-Based approach in which selected community residents compose a ranked list of households from poorest to richest whilst in a group meeting. Ideally these meetings are led by trained facilitators who help classify what it means to be poor.

A recent comparison of these two methods showed that PMTs trump the community approach on accuracy, but come up short on satisfaction ratings. The study took place in Indonesia, which is currently a leading developing country for targeted cash transfer programs, and left the government with a difficult choice to make.

Although PMTs were the most accurate, neither method was impressive in its accuracy. PMTs incorrectly classified 30 percent of households as target beneficiaries, while the community method had a 33 percent inaccuracy level.  This difference is small but significant; however, according to the authors, not significant enough to affect poverty rates if one method was chosen over the other.

Interestingly, the major method discrepancy lies in community satisfaction. Despite the community approach being less accurate, it is considered significantly more satisfactory by the targeted community than the PMT approach. Furthermore, it is more accurate than the PMT approach at identifying the households that feel the poorest (even though they are not necessarily the financially poorest).

This difference has been attributed to a community view of poverty that not only considers consumption but also earning potential and vulnerability. On this basis, communities repeatedly select widowed and less educated households as the poorest with little regard for their actual daily consumption level, while PMTs do not take such social-demographic factors into account.

As mentioned above, the households that feel poor often have a biased subjective view of their standing on the poverty scale, which is in line with that of the community but not PMTs. This suggests that there may be something valuable in the community’s alternative perceptions of poverty in spite of their supposed inaccuracy.

Taken from Figure 2 of reference below.

Nevertheless, governments wanting to implement social safety net programs for the poor are left with two major choices. Firstly, do they value accuracy above satisfaction despite the fact that the small increase in accuracy will not have substantial effects on the nation’s poverty levels? And secondly, do they just want to help those who are poor (even if they don’t feel poor) or is it also of utmost importance to help those who feel poor and vulnerable?

The logical solution would be to create a hybrid method, using a mix of the PMT and community methods. However, when the researchers of the experiment outlined above tried this they received a disappointing response. The hybrid method was no more accurate than the community method and was rated as significantly less satisfactory by the community.

This leaves researchers with the ongoing challenge of finding improved methods of identifying the poor and governments with the difficult conundrum of deciding between satisfaction, accuracy, and perceptions of poverty. In old economic theory accuracy would find itself in the lead; however, with the emerging value of psychology in development issues we may see the tables turn in the other direction.

Consideration of psychology over and above economics has many potential benefits in this instance. Through prizing community satisfaction above all else, potential implementation conflicts are eased and faith in the aid-giving institution is reinforced. Moreover, the idea that maybe there is something behind community satisfaction, and that perhaps regional knowledge about which households will most benefit from aid programs is more accurate than any PMT, should not be disregarded without a thought.

How do you think people should be selected for aid programs?

Mieke Dale-Harris is working as an intern at the Institute of Advanced Development Studies (INESAD), La Paz, Bolivia. She is a psychology graduate from Goldsmiths University of London.

Like this article? Be sure to sign up at the top of this page to receive future articles directly to your inbox.

For your reference:

Alatas, V., Banerjee, A. V., Hanna, R., Olken, B. A. and Tobias, J. (2013) Involving communities in identifying the poor. Abdul Latif Jameel, Poverty Action Lab.